ggplot2 redux

random things I like and other stuff

To cover

• facet_wrap, facet_grid
• themes, preset and custom
• ggmap
• ggally

What can you do w/ ggplot2 that you can’t do w/ base functions?

---

R Code Chunk Example

• Facetting is one of the more powerful aspects of ggplot2

• plot y vs x (or just x) by z, where z is some categorical variable

• facet_grid or facet_wrap

data(diamonds)
head(diamonds)

##   carat       cut color clarity depth table price    x    y    z
## 1  0.23     Ideal     E     SI2  61.5    55   326 3.95 3.98 2.43
## 2  0.21   Premium     E     SI1  59.8    61   326 3.89 3.84 2.31
## 3  0.23      Good     E     VS1  56.9    65   327 4.05 4.07 2.31
## 4  0.29   Premium     I     VS2  62.4    58   334 4.20 4.23 2.63
## 5  0.31      Good     J     SI2  63.3    58   335 4.34 4.35 2.75
## 6  0.24 Very Good     J    VVS2  62.8    57   336 3.94 3.96 2.48

—&twocol w1:50% w2:50%

Simple scatterplot

***=left

p1 <- ggplot(diamonds, aes(x = carat, 
        y = price)) +
    geom_point()
p1

***=right

—&twocol w1:50% w2:50%

Simple scatterplot with facet_wrap

***=left

p1 + facet_wrap(~color)

A simple scatterplot of diamond price by carat, facet_wrap by color

***=right

—&twocol w1:50% w2:50%

Simple scatterplot with facet_grid

***=left

p1 + facet_grid(~color)

A simple scatterplot of diamond price by carat, facet_grid by color

What’s the difference?

***=right

---

facet_wrap vs facet_grid

• more important for multiple facet variables
• facet_wrap always has one horizontal facet label on the top

p1 + facet_wrap(~ cut + color)

---

facet_wrap vs facet_grid

• more important for multiple facet variables
• facet_grid can have both horizontal, vertical facet labels

p1 + facet_grid(cut ~ color)

---

facet_wrap vs facet_grid

• order of variables affects position of facets
• facet_wrap orders facets by position in the call

p1 + facet_wrap(~ color + cut) # same as facet_wrap(color ~ cut)

---

facet_wrap vs facet_grid

• order of variables affects position of facets
• facet_wrap orders facets by position in the call

p1 + facet_wrap(~ cut + color) # same as facet_wrap(cut ~ color)

---

facet_wrap vs facet_grid

• order of variables affects position of facets
• facet_grid orders vertical/horizontal facets by left/right of tilde

p1 + facet_grid(cut ~ color) # not the same as facet_grid(~ cut + color)

---

facet_wrap vs facet_grid

• order of variables affects position of facets
• facet_grid orders vertical/horizontal facets by left/right of tilde

p1 + facet_grid(color ~ cut) # not the same as facet_grid(~ color + cut)

---

facet_wrap vs facet_grid

• both use the scales argument for axes, otherwise fixed

p1 + facet_wrap(~ color + cut, scales = 'free') # or 'free_x', 'free_y' 

---

facet_wrap vs facet_grid

• facet_grid treats scales differently

p1 + facet_grid(color ~ cut, scales = 'free') # or 'free_x', 'free_y' 

---

facet_wrap vs facet_grid

• facet_wrap uses the ncol argument

p1 + facet_wrap(~ color + cut, ncol = 7)

---

facet_wrap vs facet_grid

• facet_grid always creates a symmetrical plot

p1 + facet_grid(~ color + cut)

---

A huge advantage of facets…

• Very easy to quickly evaluate a variable by multiple categories
• Adding some fake variables to diamonds…

diamonds$fake1 <- sample(c('A', 'B'), nrow(diamonds), replace = T)
diamonds$fake2 <- sample(c('C', 'D'), nrow(diamonds), replace = T)
diamonds$fake3 <- sample(c('E', 'F'), nrow(diamonds), replace = T)
diamonds$fake4 <- sample(c('G', 'H'), nrow(diamonds), replace = T)
head(diamonds[, grep('fake', names(diamonds))])

##   fake1 fake2 fake3 fake4
## 1     B     C     F     G
## 2     B     D     F     H
## 3     A     D     F     G
## 4     A     D     F     H
## 5     A     D     E     H
## 6     A     D     E     G

---

A huge advantage of facets…

p1 + facet_grid(fake1 + fake2 ~ fake3 + fake4)

---

Facet summary

facet_wrap and facet_grid accomplish similar tasks, with slight differences

facet_wrap - only horizontal facet labels - not symmetric, uses ncol - scales affect all facets

facet_grid - horizontal/vertical facet labels - always symmetric - scales only affect outer facets

• Not apparent why you would use one over the other…

—&twocol w1:50% w2:50%

Themes

***=left

• ggplot2 creates plots with a given theme, default is theme_grey()

data(iris)
p2 <- ggplot(iris, aes(x = Sepal.Length, 
        y = Sepal.Width, 
        colour = Species)) +
    geom_point()
p2

***=right

—&twocol w1:50% w2:50%

Themes

***=left

• Other pre-loaded themes are theme_bw() and…

p2 + theme_bw() 

***=right

—&twocol w1:50% w2:50%

Themes

***=left

• theme_classic()

p2 + theme_classic()    

***=right

---

Themes

• Themes are simply complete calls to the ggplot2::theme function
• A custom theme is easily made

ugly_theme <- theme(
    panel.background = element_rect(fill = "green"), 
    axis.line = element_line(size = 3, colour = "red", linetype = "dotted"),
    axis.text = element_text(colour = "blue"),
    axis.ticks.length = unit(.85, "cm")
    )

• Complete list of theme options here

—&twocol w1:50% w2:50%

Themes

***=left

• Great success!

p2 + ugly_theme

***=right

---

Themes

• The user-defined theme can also be set as default by updating an existing theme

ugly_default <- function(){
    theme_grey() %+replace%
    theme(
        panel.background = element_rect(fill = "green"), 
        axis.line = element_line(size = 3, colour = "red", linetype = "dotted"),
        axis.text = element_text(colour = "blue"),
        axis.ticks.length = unit(.85, "cm")
        )
    }
theme_set(ugly_default())

p2

---

Themes

• The ggthemes library provides additional themes

library(devtools)
install_github('ggthemes', username = 'jrnold')
library(ggthemes)   

• Check the repo on Github for more info

—&twocol w1:50% w2:50%

Themes

***=left

• The ggthemes library provides additional themes
• Wall Street Journal theme

p2 + theme_wsj()

***=right

—&twocol w1:50% w2:50%

Themes

***=left

• The ggthemes library provides additional themes
• Google docs theme

p2 + theme_gdocs()

***=right

—&twocol w1:50% w2:50%

Themes

***=left

• The ggthemes library provides additional themes
• Even this old school Excel theme…

p2 + theme_excel()

***=right

---

Easy mapping with ggmap

The basic idea of ggmap is to take a downloaded map image, plot it as a context layer using ggplot2, and then plot additional content layers of data, statistics, or models on top of the map.

Kahle and Wickham 2011

---

Easy mapping with ggmap

• install/load ggmap
• download the images and format for plotting, done with get_map

install.packages('ggmap')
library(ggmap)

# get map by location
loc <- 'Environmental Protection Agency, 1 Sabine Drive, Gulf Breeze, FL'
my_map <- get_map(
    location = loc, 
    source = 'google', 
    maptype = 'terrain', 
    zoom = 13
    )

—&twocol w1:50% w2:50%

Easy mapping with ggmap

***=left

• plot with ggmap

ggmap(my_map, extent = 'panel')

***=right

—&twocol w1:50% w2:50%

Easy mapping with ggmap

***=left

• Now add some regular ggplot2 content layers

pts <- data.frame(
    lon = c(-87.1930, -87.2050, -87.1571),
    lat = c(30.3473, 30.3406, 30.3380),
    lab = c('Site 1', 'Site 2', 'Home')
    )
ggmap(my_map, extent = 'panel',
    base_layer = ggplot(pts, 
            aes(x = lon, y = lat))) +
        geom_text(aes(label = lab))

***=right

—&twocol w1:50% w2:50%

Easy mapping with ggmap

***=left

• Additional map types
• See documentation for full list of options

my_map <- get_map(
    location = loc, 
    source = 'google', 
    maptype = 'satellite', 
    zoom = 13
    )
ggmap(my_map, extent = 'panel')

***=right

—&twocol w1:50% w2:50%

GGally

***=left

• A helper to ggplot2… contains templates for different plots to be combined into a plot matrix, a parallel coordinate plot function, as well as a function for making a network plot, on CRAN
• The generalized pairs plot is a plot matrix that builds on the standard pairs plot

data(tips, package = "reshape2")
pairs(tips[, 1:4])

***=right

—&twocol w1:50% w2:50%

GGally

***=left

• Pairs plots are inadequate for exploratory analysis of variables that are a mix of quantitative and categorical information
• ggpairs provides a plot matrix of mosaic tiles that describe data of different categories using a ggplot2 framework

install.packages('GGally')
ggpairs(tips[, 1:4])

***=right

## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.

—&twocol w1:50% w2:50%

GGally

***=left

• Information above/below diagonal is not redundant
• quantitative-quantitative: scatterplot
• quantitative-categorical: boxplots
• categorical-categorical: conditional barplots

install.packages('GGally')
ggpairs(tips[, 1:4])

***=right

## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.
## stat_bin: binwidth defaulted to range/30. Use 'binwidth = x' to adjust this.

—&twocol w1:50% w2:50%

GGally

***=left

• Defaults can be customized

install.packages('GGally')
ggpairs(
  tips[,1:4],
  upper = list(continuous = "density", 
    combo = "box"),
  lower = list(continuous = "points", 
    combo = "dot")
    )

***=right

—&twocol w1:50% w2:50%

GGally

***=left

• Other plots… static parallel coordinate plots.

ggparcoord(data = iris, columns = 1:4, 
    groupColumn = 5, 
    order = "anyClass")

***=right

—&twocol w1:50% w2:50%

GGally

***=left

• Other plots… network plots.

library(sna)

url = url("http://networkdata.ics.uci.edu/
    netdata/data/cities.RData")
print(load(url)); close(url)
# plot cities, firms and law firms
type = cities %v% "type"
type = ifelse(grepl("City|Law", type), 
    gsub("I+", "", type), "Firm")
ggnet(cities, mode = "kamadakawai", 
    alpha = .5, node.group = type, 
    label.nodes = c("Paris", "Beijing", 
        "Chicago"), 
    color = "darkred")

***=right

---

What does ggplot2 offer that isn’t in base?

• Many geoms

ls(pattern = '^geom_', env = as.environment('package:ggplot2'))

##  [1] "geom_abline"     "geom_area"       "geom_bar"       
##  [4] "geom_bin2d"      "geom_blank"      "geom_boxplot"   
##  [7] "geom_contour"    "geom_crossbar"   "geom_density"   
## [10] "geom_density2d"  "geom_dotplot"    "geom_errorbar"  
## [13] "geom_errorbarh"  "geom_freqpoly"   "geom_hex"       
## [16] "geom_histogram"  "geom_hline"      "geom_jitter"    
## [19] "geom_line"       "geom_linerange"  "geom_map"       
## [22] "geom_path"       "geom_point"      "geom_pointrange"
## [25] "geom_polygon"    "geom_quantile"   "geom_raster"    
## [28] "geom_rect"       "geom_ribbon"     "geom_rug"       
## [31] "geom_segment"    "geom_smooth"     "geom_step"      
## [34] "geom_text"       "geom_tile"       "geom_violin"    
## [37] "geom_vline"

---

What does ggplot2 offer that isn’t in base?

• Easy facetting, pre-loaded and customized themes, spatial data, pairs plots…

• See online documentation for additional functionality

• Presentation materials available here

• EPA, NHEERL slidify template from Jeff’s repo